THE CONSTANT ERROR IN AMPLITUDE DISCRIMINATION AND THE INTER-STIMULUS INTERVAL: A RE-EXAMINATION.

The constant error or time error observed in psychophysical discrimination experiments has been a source for conjectures about the judgment process, the suggestion being that the nature of the errors indicates something significant about the underlying decision mechanism. The classical experiment for this purpose shows how the constant error or response bias changes as the inter-stimulus interval varies. This relationship is examined for the case of amplitude discrimination by using an XAB paradigm which switches between several amplitude levels randomly from trial to trial. Data from three very experienced observers indicate that with this procedure, variations in the inter-stimulus interval do not affect the response bias, but the discriminability decreases regularly with increasing intervals. A signal detection-type model is developed for the experiment which is used to describe two alternative interpretations for the effect of the inter-stimulus interval. The first interpretation follows the classical fading trace notion, while the second postulates a random walk for the psychological representation of stimulus X. It is shown that neither hypothesis alone is sufficient to account for the discriminability changes that occur with the inter-stimulus interval effects characteristic to each interpretation appear in the data. Author